This is my first post on this forum so best I introduce myself. A significant part of my career has been involved with Satcom ( 18 years with geosynchronous SV's) and more recently with GPS and product performance testing. Much of what has been posted is on point however the reasoning behind it borders conjecture. The GPS system send down GPS packet data at a predetermined 1hz rate (1 per second). We have no control over that. Some products record at half that rate only recording every other GPS point. Distance accuracy is only one measure of an accurate GPS workout. TTFF ( time to acquire a satellite fix) is very important. Positional accuracy at TTFF, Altitude accuracy at TTFF etc etc. Some products perform "smoothing" which will provide a better looking graph but affects accuracy. Some manufacturers include a "fudge factor" to compensate for a poorly tuned antenna/reciever pairing. It's great when the run comes up short but will over report when the distance actually comes in accurate. I do side by side by side testing with significant controls and very accurate reference equipment capable of decimeter accuracy. GPS should NEVER report long, NEVER. Tendency is to miss points along the way not generate distance. Sufficient algorithms are in the firmware of the big player GPS watches that handle arm movement artifacts to varying degrees. The latest technology is to get extended emphemeris from cell tower broadcasts or from server generated sites when you offload your workout. The emphemeris could be good(accurate) for up to 7 days in some instances. The GPS system has so many variables affecting any given workout, some of which were mentioned above, like tall buildings( multipath), wet canopy attenuation,time of day, atmospheric interference, etc etc.
I'll stop here since I could go on and on about anything relating to GPS and especially about accuracy. I would entertain any specific questions and anawer as time allows. I'm looking forward to some interesting reading.
Robert